POI (Point-of-Interest) Data knowledge base​

Welcome to our Quadrant POI Data Knowledge Base! Through these chapters, we will learn about what POI data is, why it is useful, and the mechanics behind its collection and real-world usage.

There has been a seismic shift in how consumers interact with businesses and locations around them. Hailing a cab on the street has been replaced by ride-share services like Uber and Grab. Instead of queuing or making appointments at their favourite restaurants, people have their food delivered to their doorstep. There is an app for everything, from buying clothes to running errands, from finding housekeeping or laundry services, and so on. It is almost unimaginable that we once lived in a world where all these services did not exist.

With the rapid increase in 'at your door' services, POI data becomes vital to efficiency. Without an accurate reference of location, delivering these services would be difficult. Uber drivers need to know the restaurant's exact location to pick up the next passenger, and it makes sense for a fast-food chain to open a new store in a growing residential area where customers are underserved.

Follow us on LinkedIn to get updates on new chapters and topics that we are constantly adding, or send us your questions to marketing@quadrant.io!

A Point-of-Interest (abbreviated POI) is a record of a place on a map that someone finds useful or interesting. A POI is typically defined by its geographical coordinates and a few additional attributes like name and category: The Fullerton, a hotel at latitude 1.286546, and longitude 103.853721, or the Empire State Building at latitude 40.748817, and longitude -73.985428.  are good examples of POIs. 

Where does POI data come from? How do data providers and vendors generate or acquire POI data? Let us look at some of the most popular POI data collection methods.

Data extraction from web sources

The most rudimentary way of acquiring data is the automatic extraction of POIs from web sources like Google Maps, OpenStreetMap, etc. Some companies use web scraping tools to export POI information directly into a file or database. While sites like OpenStreetMap allow the extraction of geospatial data points, others are actively preventing mass scraping with stopgap measures like IP blocking. Getting around these might not be illegal, but it is discouraged. Overall, this method is time-consuming and labour intensive, and it is not ideal for large-scale POI projects that need a larger volume of data. The data acquired from these sources also requires intensive data preparation to be used for analysis.

Location engines like Google Maps also heavily rely on businesses and places supplying their own information. If they do not voluntarily update these data points, information can become outdated and introduce inconsistency and inaccuracy into datasets.

User Generated

In the past decade, there has been a surge in the use of social media and thus in user-generated content, including location data. Many businesses rely on user-generated location data or purchase data from vendors who run applications that collect such data. Remember your last check-in on your favourite social media app? You probably added a new location or verified one for a location data collector.

Location data collectors record locations pinned by users and create their own POI records based on that information. However, geographic data provided by a user does not guarantee accuracy, as it is dependent on the hardware, the location permissions of the application, and precision settings. Unintentional mistakes can also deteriorate the quality.

Government Directories

Most government bodies require businesses to provide their commercial location as part of their business registration process. These postal addresses can form a POI dataset. Many countries make this data publicly available. For example, the Accounting and Corporate Authority (ACRA) of Singapore makes the historical and current data of 1.5 million companies available for public use, research, or application development purposes under their open data initiative.

However, businesses close, grow and move offices or operate at a different location than their officially registered address. Since they may not proactively update this information with the government body, these data can become outdated and inaccurate over a period of time.

Manual Verification

Some POI data providers hire or contract personnel to manually maintain their POI database: walking the city with a smartphone running a purpose-built app and adding new locations or verifying existing ones. Compared to other methods, this process guarantees accuracy and promises a healthy stream of POI data.

This method of data collection also avoids infringement on users’ privacy and sale of their location data. These purpose-built apps do not store, collect, or share any data other than the physical location (without tying context back to an actual human being and their mobile device).

POIs can be placed on a map using a variety of formats, all coming with a unique way of data representation and granularity. Using these formats, POI attributes can be used to specify a place on a map and define its spatial relationship with places around it.

Coordinates

Commonly expressed in latitude and longitude (Lat/Long), the geographic positions of a location are called coordinates. A POI can be identified with data points as minimal as the latitude and longitude. These data can be procured from satellite-based mapping services or collated using GPS-enabled devices (such as smartphones, fitness trackers, tablets, etc.). Using multiple coordinates, it is possible to create a larger virtual perimeter for a real-world geographic area known as Lat/Long Boundary or Geofence. A boundary can be radial or in the shape of a polygon formed by connecting multiple Lat/Long points. These boundaries help businesses identify and assess POIs within a specific area to draw trends and patterns.

Postal Codes

Another common method to represent a POI is its physical address, which usually includes a government-assigned Pin, Postal, or Zip Code. Initially aimed at helping mail carriers sort and deliver mail efficiently, the postal code system became widely used to determine the location of a place of interest. However, there is no global standard for postal codes. In some countries, like Singapore, a postal code can identify an individual housing block. On the other end, Hong Kong, a territory comparable in size and population, does not use the postal code system at all. Most places fall somewhere in between, where a postal code is not enough to pinpoint an individual Point-of-Interest but can be used as an “official geofence” in data analysis.

Geohash

Invented by Gustavo Niemeyer, Geohash is a geocoding system that allows the expression of a location anywhere in the world using an alphanumeric string. Geohash is a unique string derived by encoding and reducing the two-dimensional geographic coordinates (latitude and longitude) into a string of digits and letters. A Geohash can be as vague or accurate as needed depending on the length of the string. A Geohash uses grids for spatial indexing where the world is recursively divided into small grids, with each added grid introducing an extra level of accuracy. One of the primary benefits of a Geohash is that it is excellent for precisely pinpointing a POI. By dividing a larger area into grids, you can eliminate most unwanted areas upfront and only focus on the square where your potential targets are in. It also allows for faster geofencing than the lat/long method, saving time and money.

H3

H3 is a hierarchical geospatial indexing system much like Geohash. However, it uses hexagonal grids instead of rectangular grids. H3 was developed by the popular ride-sharing app Uber to optimise ride pricing and dispatch by visualising and exploring spatial data. The H3 grid system was open-sourced in early 2020 and is slowly gaining popularity. The H3 system works best for efficient radial lookups. Which is why H3 is best for calculating distance between two places and allows circular geofences. 

Information about a specific location or a collection of similar locations can help businesses drive better decisions. Combined with additional contextual parameters such as human mobility, sociology, dynamics of an area, and more, POI can be used to build meaningful information structures that enable robust analysis and planning. 

An ideal POI database evolves with changes happening in the physical world. New businesses open, existing ones move or close, neighbourhoods expand, new parking facilities become available in high-density areas (where they are most needed), and more. Furthermore, urban spaces are constantly changing. Recently, our world was greatly impacted by the COVID-19 pandemic, as many businesses went out of business and closed. All these changes in the physical world need to be documented to keep our complex digital economy going, and POI databases do just that.

But where do you get POI data that is fresh, accurate, and reliable? There are numerous POI data providers who map the world using various methods, and enrich it by adding new places and removing outdated information.

• Define your goals and challenges
  
  
• Identify the category of location data you need.
  
  
• Determine the frequency of updates (daily, monthly, annual)
  
  
• Align accuracy with your use case (What happens if the accuracy is off by 100 meters, how does it affect your business?) 

Here are the key parameters to assess and choose the right POI data partner: 

Whether you are getting POI data short-term batch processing or long term, recurring analysis asses the method of delivery best suited to your business needs and resource availability.

Most existing, widely available POI databases are quite inaccurate, and consumers often face mismatched information based on exact locations. Accurate POI information is valuable now more than ever, as the COVID-19 pandemic has rewritten the maps of businesses. Some companies were forced to shut down, and many moved online to sustain themselves. As the economy recovers and new POIs (Points-of-Interest) appear, companies will need accurate data to function, especially those in ad-tech, food and dining, communications, and navigation. To reach accurate outcomes, new domains in data science require superior quality POI input that companies like Google and other non-caching APIs do not offer.

To solve this problem, Quadrant has come up with a diligent, human based approach to collecting POI data. Quadrant POI Data is powered by Geolancer - an industry-leading manual verification project. How this works is, actual human verifiers, equipped with a smartphone run our proprietary app and manually add/verify POI data points, ensuring accuracy and authenticity. This unique process solves most of the problems plaguing other POI data sets, including user errors, GPS inaccuracies, and outdated information. The platform can be used to create bespoke data sets based on client requirements (for a specific area, for a category, or brand) or to verify a company’s existing POI data.

Our goal is to create solutions for people and the businesses that serve them. With our POI offerings, our customers will finally be able to apply analytical models against large, accurate POI datasets, deriving meaningful and actionable insights. With the increasing digital transformation of businesses, there is an enormous market need for quality POI data. Using Geolancer, we will be able to offer authentic, accurate, and up-to-date POI data to companies in the ride-sharing, real estate, e-commerce, delivery and logistics and other location powered industries.

POI Data combined with our expertise in mobile location data solution can provide deeper insights on the behaviour of our customers’ target audience making us a holistic data partner for them.